Major network problem

[Andyhall]

We have a Win 2003 network and are having major network problems since the start of term. We have traffic bottlenecks which seem to cause switches on the network to fail. This bottleneck seems to go back to the previous switch as well. It has now happened three times in as many days to different parts of the network. In each case different switches have failed. The 3Com ones have been reset to their old cofiguration PI address etc and sem to work OK again.

We had this problem at the same time last year but only on one part of the network. It only seems to occur at the start of the school year

We have considered several options:

There are only two things installed over the summer. We have set up a thin client network but this was not in use on when the system first went down on the first day back. We have also just set up PARS - a lesson monitoring system, which links to SIMS. Could this be the problem?

Maybe there is something on our network that has been brought in from outside on a laptop or other device?

 

We are a bit stuck as to what to do next.

1) Our own fixes – trying this but we are still having problems.

 

2) Get some outside help from a support company that we already deal with

3) Install some network monitoring software to monitor network. This may cost and would not be retrospective. We have been told about various network monitoring problems by various people but are not sure what to use!

 

HELP!

[SYNACK]

Possibly a loopback that someone has created accidentally on your network of a faulty network card somewhere. These threads may help:

 

http://www.edugeek.net/forums/networks/19195-solved-entire-network-down-100-network-utilization-please-help.html

http://www.edugeek.net/forums/networks/23282-arp-flood.html

http://www.edugeek.net/forums/networks/6306-stop-looping-effects-caused-interconnection.html

[djdohboy]

One problem I have come across between switches is that auto negotiation fails spectacularly sometimes and they cannot decide between them what speed they should be running at, the way to get round this is to force the link ports on both ends to the same speed (ie 1000mbps Full duplex).

 

This has worked well for me in the past

[Andyhall]

Thanks for the help and links, which I found really useful

[zag]

We had the same kind of error on the first day back. It turned out our chemistry department had plugged in a home network hub and looped a cable on it.

 

Unplugged it and the network sprung to life again!

[MSGeek]

It looks like a spanning tree problem, but might not be a broadcast storm, if it was it would be permanent until the loop is removed, it is worth investigating that avenue. To me it looks like someone is plugging their own switch into the network causing the root bridge to be recalculated, or even the addition of the switch is making the diameter of your network exceed the 7 hop limit imposed by 802.1D, unless you are running 802.1w which is then a 20 hop limit. but in any event I would do the following:

1. Statically set the speed and duplex of all links.

2. If the 3com support UDLD then turn it on on all up links copper and fibre

3. Statically set the root bridge

4. Enable "bpduguard" on all access links

 

I am talking from a HP/Cisco world and don't know if 3com support these features

 

Hope this helps